Sampling for Reliability Assessment Using Weighted Prior Data.
Abstract
The report presents the details of a procedure for determining a sample size, n, from a population of complex systems for reliability assessment when it is planned to combine the new data with past experimental or subjective information. The method assumes a Bayesian rationale where the parameter of interest, reliability, is assumed to be a random variable. The criterion for choice is that the sample size be sufficiently large to yield a posterior distribution for reliability with a specified minimum mode and lower 100 (1-alpha)% confidence limit for a specified alpha when it is assumed that no failures will occur. The mechanism for data combination employs 'weighting factors' defined as a function of the parameters of the prior distribution and likelihood function. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1971
- Accession Number
- AD0725568
Entities
People
- Jeffrey P. Schneider
- John G. Mardo
Organizations
- Picatinny Arsenal